Brittman: still there.

 
Losh: A ton of **** changed in the optimizer

Brittman: Scott0_: you moved the data files, not the binaries. you have no idea if the QEP changed because you haven’t run explain.

Brittman: Scott0_: query execution plan.

Sarchet: Why would this not be in the upgrade docs?

Brittman: Scott0_: if the plan is exactly the same, same number of rows read in handler status, and you were i/o bound, maybe you had a fast raid 10 before and now on some slow raid 5

Brittman: Scott0_: you dont have to upgrade to see this

Rius: If a newer version changes something that could break a db being moved over, shouldn’t that be do***ented?

Dykas: Im seeing different row count on the explain

Schonack: And im not sure what to make of that

Velis: They are estimates only. differences are expected.

Ganz: Wait I think the old db is mising some recent data

Outland: So how can I get the same application and database performing how it was on 5.1?

Houskeeper: So I upgraded for nothing?

Bolar: Everyone said to upgrade to 5.5

Brittman: Scott0_: can you share the two explains, and how long each query takes?

Carron: Brittman: I was mistaken about the explains

Edds: Brittman: I was selecting a dataset that was in a date range that only the new server had

Brittman: Scott0_: you captured the exact query from general log, made sure it was slower in 5.5 than 5.1, then compared explains?

Brittman: Scott0_: can you share the two explains, and how long each query takes?

Hattaway: I did modify the date range

Evins: So that both servers would have the same data

Grimmius: Im afraid this is such an ugly and big query that its going to convolute things

Shorthair: As well as expose some private info

Brittman: Just the explain line

Alphonso: The query or the result?

Brittman: You said it is a single table and a single index didn’t you?

Brittman: Just the result, and the time to actually execute the query, as a start

Khlok: Yes but the query is funky

Magri: Its a ton of self left joins

Brittman: Scott0_: did the join order change?

Danfield: Scott0_: I’m not sure you know what EXPLAIN is

Nania: The explain contains some names which id rather not display publicly 😛

Brittman: No one could stop you

Oestreich: Im not sure how to without destroying the pretty formatting of the explain

Gustason: Finally fixed the blank grid bug for workbench. need to compile version 6.2 manually and apply a glib patch https://bugs.launchpad.net/ubuntu/+source/mysql-workbench/+bug/1376154/comments/7

Majeau: Each joined table is aliased as a different name

Brittman: Scott0_: does explain itself take a long time?

Degiacomo: So no big difference there

Brittman: Try lowering optimizer search depth

Pritt: Which made me think of maybe IO since it may be the time of delivery of the data

Coffer: Maybe the new KvM VPS has horrible IO

Spurgeon: Scott0_: See http://google.com/search?btnG=1&q=bonnie+io

Savell: Im guessing this is something I have to shutdown all services to run

Brittman: Not if production is faster than the old box even with side load

Redway: Faster as in IO or CPU?

Sibilia: Brittman: I didn’t understand your last comment

Krawitz: New server seems to be faster

Wildey: I even started the old server first

Brittman: Then you did understand my comment

Chock: Yeah both are runing the same load

Salerno: Brittman: still there?